What is a deepfake?
In one line
Ages 8 to 10: A deepfake is a picture, video or voice that a computer faked to look like a real person.
Ages 11 to 13: A deepfake is made by AI that learned patterns from many real examples and uses them to make new, believable fakes.
Ages 14 to 18: A deepfake is synthetic media of a real person; because detection is unreliable, checking the source and the story matters more than inspecting pixels.
A deepfake is a fake picture, video or voice recording made with AI so that it looks or sounds like a real person saying or doing something they never did. Deepfakes can look very real. If you see one about someone you know, don't share it, and tell a trusted adult.
Last reviewed: 20 October 2026 7 min read
The idea in one picture
Think of a very skilled copy artist. Show them hundreds of photos of one face, and they can draw that face smiling, frowning, or in a place it has never been. A deepfake is the computer version: AI studies many real examples of someone's face or voice, then makes new images, video or sound that were never recorded.
Worked example. A short clip of someone's voice is online. A voice-copying program uses it to make new speech in that voice. The new audio says words the person never said. Nothing in the sound alone proves it is fake, which is why the check has to happen another way, such as calling the person on a number you already know.
How it works, step by step
- Someone collects real examples of a person: photos, video or a voice clip.
- An AI system that has learned patterns from huge numbers of images or recordings uses those examples.
- It generates new pictures, video frames or sound that match the person's look or voice.
- The pieces are put together into an image, clip or audio message.
- The fake is shared, often with a story that makes people react quickly.
A little deeper (ages 11 to 13)
Deepfakes come from generative AI, the kind of AI that makes new content instead of sorting or labeling it. It works from patterns learned in its training data. People use the same tools for harmless things, such as film effects or a clearly labeled joke. A deepfake becomes harmful when it is used to fool people, embarrass someone, spread false news or trick someone into sending money or secrets.
Going further (ages 14 to 18)
Researchers call this kind of content synthetic media. Deepfakes make it harder to tell human-made content from AI-made content, and as tools improve, small visual "tells" become less reliable. So the stronger habit is checking provenance: where something came from, who first posted it, and whether trusted sources report the same thing. Some tools add watermarks or record a file's origin in its metadata, which can help, but missing labels do not prove something is real. A deepfake can also cause harm by making people doubt true evidence.
Common mix-ups
- Mix-up: "I can always spot one."Fact: Some deepfakes are very realistic. Check the source and the story, not only the picture.
- Mix-up: "If it's a joke, it's harmless."Fact: A fake that embarrasses a real person can hurt them, even as a joke.
- Mix-up: "A familiar voice proves it's them."Fact: Voices can be copied from a short clip. Call back on a number you know.
- Mix-up: "Every edited photo is a deepfake."Fact: Brightening or cropping a photo is ordinary editing. A deepfake makes a real person appear to say or do something they did not.
What to do about it
- Pause before reacting. Urgent messages asking for money, passwords or secrets are a warning sign.
- Don't share or forward it, even to say "look how fake this is." Sharing spreads it further.
- Tell a trusted adult at home or at school. You are not in trouble for seeing something.
- Report it on the app or site where you saw it, using its report button.
- Check a voice message by calling the person back on a number you already know, as the US Federal Trade Commission advises.
- For grown-ups: If an explicit image of someone under 18 is shared, real or AI-made, anyone, including the young person, can report it to the National Center for Missing & Exploited Children's CyberTipline at report.cybertip.org or its 24-hour hotline, 1-800-843-5678. NCMEC's Take It Down service can help a young person request removal of an explicit image taken before they turned 18 from participating platforms; it works from a digital fingerprint, called a hash, so the image never leaves the device. A trusted adult can help them, or make a CyberTipline report (sources: NCMEC, "Generative AI (GAI)"; NCMEC, "Take It Down"; NCMEC, "Take It Down" FAQ).
This is general safety guidance, not legal advice.
Try it (no device, 5 minutes)
Real or remade? One person tells two short stories about their day; one is true and one is invented but full of real details. Others ask up to three questions to check. Afterwards, talk about it: Which details made the made-up story believable? What question helped most? That is the same habit that works for deepfakes: check the story, not just how real it looks.
Sources
- National Institute of Standards and Technology, "Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile" (NIST AI 600-1)
- US Federal Trade Commission, "Scammers use AI to enhance their family emergency schemes"
- US Federal Trade Commission, "Did someone share an explicit image of your child online? Take these steps"
- National Center for Missing & Exploited Children, "Generative AI (GAI)"
- National Center for Missing & Exploited Children, "Take It Down"
- National Center for Missing & Exploited Children, "Take It Down" FAQ
- National Center for Missing & Exploited Children, "CyberTipline"
Quick check
What makes something a deepfake rather than an ordinary edit?
Answer: It makes a real person seem to say or do something they never did.
It makes a real person seem to say or do something they never did.You get a voice message from a relative asking for money urgently. What do you do?
Answer: Don't reply or pay; tell an adult and call the relative back on a number you already know.
Don't reply or pay; tell an adult and call the relative back on a number you already know.(Ages 14 to 18) Why is checking the source more reliable than looking for glitches?
Answer: As tools improve, visual flaws disappear, but where content came from can still be checked.
As tools improve, visual flaws disappear, but where content came from can still be checked.Questions
A deepfake is a fake picture, video or voice recording made with AI to look or sound like a real person. It can show someone saying or doing something they never did.
Often you cannot tell from the image or sound alone, because good deepfakes look real. Check where it came from, whether trusted sources report the same thing, and, for a voice message, call the person back on a number you already know.
The same tools are used for film effects and clearly labeled jokes. A deepfake is harmful when it is used to fool people, embarrass someone, spread false news or trick someone into sending money or secrets.
Don't share or forward it, and tell a trusted adult at home or at school. Report it on the app or site where it appeared.
Report it to the National Center for Missing & Exploited Children's CyberTipline and ask the platform to take it down. NCMEC's Take It Down service can help a young person request removal of an explicit image taken before they turned 18 from participating platforms. A trusted adult can help them, or make a CyberTipline report.
Mechna Academy is a city of machines for curious children aged 8 to 18, opening first to ages 8 to 10. In each mission, a child guesses how a machine decides, then opens its Peek Hatch to check, changes one thing and tests it on a case it has never seen. Children learn how AI works this way, and they never chat with an AI.
For grown-ups: join the waitlistPlaces are limited. The waitlist is for parents and guardians 18 and older.